Chap 5

Layout Planning Models

and Design Algorithms
 Lecture outline
1. Introduction
2. Basic layout types
3. Layout procedures
4. Algorithmic approaches
5. Department shapes and mail aisles
6. Simulated annealing and genetic algorithms
7. Multi-floor facility layout
8. Commercial facility layout packages
9. The impact of change
10.Developing layout alternatives
2
I. Introduction

• Generating layout alternatives is critical steps

– Comprehensive and creative
• Layout:
– Block: relative locations and sizes of planning
departments
– Detailed: exact location of all equipments work
benches, and storage areas within each
department.
• We will focus on block layout.

3
• Which comes first?
-Material handling or facilities layout / block or detailed layout.
• Many appear to believe that it should be the layout. Facilities layout
is considered first – overemphasis on manufacturing.
• For example:
1. If parts move from dept. A to B
- It is logical to have dept. B next to A
- Handling is minimized
2. If parts cannot move from dept. A to B directly.
- WIP storage is necessary
- Parts move from A to storage and then to B
- In such a case, A and B need not to be close to each other.
• Both facilities layout and material handling should be designed
simultaneously.
4
II. Basic layout types

Fig 06_1 shows alternative

types of layouts: a) Product
layout, b) Fixed product
layout, c) Product family
layout, d) Process layout.

(Review to chapter 3)

5
III. Layout procedures
• Many procedures are available
• Broadly classified
– Construction type layout methods
• Layout from scratch
– Improvement type layout methods
• Alternatives based on existing layout
• Apple’s plant layout procedure (read page 296-297)
• Reed’s plant layout procedure

- The most importance is the layout planning chart. (Figure 6-2)

6
Fig 06_02: Layout planning chart

7
• Muther’s Systematic Layout Planning (SLP) procedure

Fig 06_03: SLP procedure Fig 06_04: Activity relationship chart 8

Fig 06_05: Relationship diagram (RD)
Fig 06_06: Space relationship diagram

9
Fig 06_07: Alternative block layout
IV. Algorithmic Approaches
• Relative placement of departments based on
– Closeness ratings
– Material flow intensities
• SLP and other approaches discussed earlier are not format
approaches
• Algorithm classification: Based on input to the algorithms
– Only qualitative data (RD)
• Need input from several individuals. Not practical for problems with 20 or more
departments
– Only quantitative data (from-to chart)
• Predominantly used in practice. Requires more time and effort to prepare from-to chart
– Both qualitative and quantitative data
• BLOCPLAN

• Objectives:
– Minimize the sum of flow distances
• From-to matrix
– Maximize an adjacency score 10
• Relationship chart
 1. Algorithm classification,
Flow distance objective
• fij: flow from department i to j (expressed in number of
unit loads moved per unit time).
• cij: cost of moving a unit load one distance unit from
department i to j
• dij: distance from department i to j

• Note: cij: independent of the material handling equipment

used department i to j 11
 Adjacency score objective

The objective is to maximize the adjacency score; that is:

For the “normalized” adjacency score or efficiency ratio:

Normalized adjacency ratio is [0,1]. When it is 1, it means that all the departments
with a positive flow are adjacent to each other.
12
 Adjacency score objective (con’t)
• Some times it is unimportant (relationship is represented as X) to
have departments I and j adjacent to each other.
– A negative fij is used to avoid i and j adjacent to each other

– Where F and F represent the set of department pairs with positive and
negative flow values.

• Adjacency objective do not account for the distance between

departments.
– Consequently, two layouts with same score can result in different total
distance between departments.
13
Choosing an appropriate grid size is important . If the grid size is smaller
there are larger number of grids-computationally burden.

Fig 06_8: a)
Discrete vs b)
continuous
layout
representation

14
• For a rectangular department with known area
– If you know the x, y coordinates of centroid of the department and its
length along the north-south direction, it is easy to represent its exact
location and shape
• If a department is too large, it maybe represented as two departments
– But the two departments cannot be split in the final layout

15
Fig 06_09: Examples of split and unsplit departments.
2. Pairwise Exchange Method

• Improvement type layout algorithm

• Current layout has to be changed
– New machine is includes
– Space required for storage has to be altered
– Existing layout is poor
–…
• Can be used for both adjacency based or
distance based objectives.

16
 • Assume all the departments are of equal size, for simplicity
Table 6-1: Material flow matrix Table 6-2: Distance matrix based on existing layout

• TC1234 = 10(1) +15(2) +20(3) + 10(1) +5(2) +5(1) = 125

a)

17
b)

c)

18
3. Graph-Based Method – a construction type.
– a construction type, with an adjacency-based objective.

Fig 06_12: Relationship chart and relatioship

diagram for graph-based example.

Fig 06_11: Adjacency graphs for alternative block layouts. 19

• Some observations:
– Adjacency score does not account for distance, nor does it account for
relationships other than those between adjacent departments.
– Dimensional specifications of departments are not considered.
– Planar graph – arcs do not intersect (adjacency graph). Non-planar graph –
relationship diagram.
– Score is very sensitive to the assignment of numerical weights in
relationship chart.

• Maximize weighted planar adjacency graph

– Approach 1:
• Start with the relationship diagram
• Prune arcs while making sure the final graph is planar
– Approach 2:
• Iteratively construct an adjacency graph via a node insertion algorithm
while retaining planarity at all times

20
• Node insertion heuristic
– Step 1: from the relationship chart, select the department pair
with the largest weight. Break ties arbitrarily.

– Step 2: select the third department to enter. Third department is

chosen based on the sum of the weights with respect to
department 3 and 4.

21
– Step 3: 4th deppartment is chosen by evaluating the value of adding one of the
unassigned depts. represented by a node on a face of the graph.
• A face of a graph is a bounded region of a graph.

• Recall: we are trying to build a planar graph

– Step 4: evaluate inserting department 5 in different faces, namely 1-2-3, 1-2-4,
1-3-4, and 2-3-4.

22
– Step 5: using the adjacency graph, draw the block layout.

Fig 06_14: Block layout from the final adjacency graph.

23
4. CRAFT (Computerized relative Allocation of Facilities Technique)
• From/to chart is the input
• Distance based objective
• Departments are not restricted to rectangular shapes
• Discrete layout
• Improvement-type layout algorithm
• Centroids for each department is calculated
• Rectilinear distance between depts. Using centroids
• All possible two-way (pairwise) or three-way department
exchanges are considered.
– Depts. with equal area
– Depts. which are adjacent to each other.
• Steepest descent procedure
• May stop at local optima
• Initial solution matters 24
 Example 6.1. Using CRAFT to improve an existing layout

Each grid is 20 x 20 sq.ft

Table 6-3: Departmental data and fr/to chart

Total area required (70,000 sq.ft) is less than available (72,000 sq.ft).
The extra space (2000 sq.ft) is modeled as a dummy department.

25
 • Distance between A and B is 6
grids
• Flow between A and B is 45 units
• A and B: 6 x 45 = 270
• Total z =2974 grids
• Note: The location of department
A (receiving) and G (shipping) is
fixed. They cannot be considered
Fig 06_15: Initial CRAFT layout and department
for two- way exchanges. centroids (z = 2974 x 20 = 59.480 units)

• Departments E and F are adjacent

• Area required for E and F are
different
• First draw a single box around E and F
• Choose the bigger dept.
• Start from left or right column to fill
• Label first 20 grids of dept F as dept E
Fig 06_16: Intermediate CRAFT layout obtained after
exchanging departments E and F (z = 2953 x 20 =59,060 units) 26
• Exchange departments B (8,000sq.ft) and C (6,000sq.ft)

Fig 06_17: Final CRAFT layout (z = 2833.5 x 20 = 56,670 units)

Fig 06_18: Final “massaged” layout obtained with CRAFT

27
 • Consider a 7*5 layout
below
Exchanging departments 2
and 4 will lead to splitting
department 2 – not
acceptable

Fig 6_19: Example to show that CRAFT may not be able to exchange two
adjacent departments that are not equal in area 28
5. MCRAFT
• Input
– Building length
– Building width
– Number of bands
– Initial layout vector
– The program calculates appropriate grid size, the number of rows and columns

29
Fig 6_21: Final MCRAFT layout ( z = 57,333.34 units)
30
6. BLOCPLAN

• Similar to MCRAFT
• Input
– From –to chart
– Relationship chart
• The number of bands is determined by the
program
• Number of bands limited to two or three
• Bands widths may vary
• Construction and improvement algorithm
31
7. MIP (Mixed Integer Programming)
Parameters:
• Bx the building length
• By the building width
• Ai the area of dept i
• Lil the lower limit on the legth of dept i
• Liu the upper limit in the legth of dept i
• Wil the lower limit on the width of dept i
• Wiu the upper limiit on the width of dept i
• M large number
Decision variables:
• I the x-coordinate of the centroid of dept i
• I the y coordinate of the centroid of dept i
• Xi’ the x coordinate of the left (west) side of dept i
• Xi” the x coordinate of the right (east) side of the dept i
• Yi’ the y coordinate of the top (north) side of dept i
• Yi”the y coordinate of the bottom (south) side of dept i
• Zijx = 1 if dept i is strictly to the east of dept j, and 0 otherwise
• Zijy = 1 if dept i is strictly to the north of dept j, and 0 otherwise 32
33
8. LOGIC

34
Example 6.6: Using LOGIC and cut-trees to construct a layout

• None of the depts are fixed

• Length = 360 ft
• Width = 200 ft

First Vertical cut

– Area of D, F, G = 36,000
– Width = 200
– Length = 36000/200 = 180

35
Horizontal cut
• AB total area = 20,000 Horizontal cut
• Length = 180 • DF total area = 24,000
• Width = 20,000/180 = 111.11 • Length = 180
• Width = 24,000/180 = 133.34

Fig 06_26: Layout obtained by horizontal and vertical cuts executed by LOGIC
36
 Fig 06_27: Cut-tree for
example 6.6
Fig 06_26 (con’t)

37
- LOGIC can also be used as an improvement algorithm
- Suppose we consider a pair-wise exchange (D and E)
- Swap D and E in the tree shown in the previous slide
- Recalculate the x & y coordinates

Fig 06_28: LOGIC layout obtained after exchanging dept D and E in fig 6.26d.
38
9. MULTIPLE (Multifloor Plant Layout Evaluation)

10 .CORELAP (Computerized Relationship Layout

Planning)

11. ALDEP (Automated Layout design Program)

39
V. Department shapes and Main Aisles

Fig 06_33: Alternative department shapes and the smallest enclosing rectangle (SER)

40
VI. Simulated Annealing and Genetic Algorithms
1. Simulated Annealing
• so the initial layout vector
• s* the “current best” layout vector (lowest cost layout)
• s the current layout vector
• s’ the candidate layout vector
•  the temperature reduction factor
• T a set of annealing schedule temperatures {t1, t2, t3, …} where ti = t0()i for all i>1
• t0 the initial temperature
• e the (fixed) epoch length
• fj(s) the objective value of the jth accepted candidate layout vector, s, in an epoch
− − e 
• f e the overall mean objective function value of an epoch [i.e., f e =  f j (s) / e
 j =1 
−
'
• f e the overall mean objective function value of all the layout vectors accepted during
the epochs previous to the current one.
• i be a threshold value used to determine whether the system is in equilibrium at
temperature i.
• M: the maximum total number of epochs to be considered
• I: a counter to record the last temperature setting that produced the current best layout
vector, s*.
• N: the maximum number of successive temperature reductions that will be performed with
no improvement in s* 41
1. Simulated Annealing procedure:

2. GA (please read in textbook) 42

VII. Multifloor Facility Layout
• CijH (CijV ) the cost of or the relative “weight” associated with, moving a unit load
one horizontal (one vertical) distance unit from department i to j
• dijH (dijV ) the horizontal (vertical) distance from department i to j

• Suppose the set of positive flows is F = {fij}. That is, fij>0 for all I, j F. Let M: the number
of positive flows in the flow matrix. And fmij: is the mth positive flow (m = 1, 2, 3, .., M)
The linear MIP model is given:

43
VIII. Commercial Packages for facilities layout
• Factory CAD, FactoryFlow, FactoryMockup and
Plant Simulation
• …

IX. The impact of change

• Volvo facilities (read textbook)
X. Developing layout alternatives
1. Exert the necessary effort to generate and evaluate the
alternatives.
2. Set a time limit and use the available time …
3. Seek many alternatives.
4. Establish a goal and layout performance metrics
5. Make liberal use of the questioning attitude
6. Don’t “fail to see the forest for the trees”
7. Think big. Then think little.
8. Don’t be conservative.
9. …
16. Do your homework.
17. Don’t overlook an improved version of the present method.
18. Think long range.